An Explainable Deep-Clustering Approach to Assess Climate Vulnerability in India

Soumyajit Basu, Fiona Arora, Aryan Nath, Karan Kumar, Lipika Dey

International Conference on Pattern Recognition and Machine Intelligence (PReMI)

Compact Private Set Intersection with Authentication and Authorization

Aryan Nath, Manya Garg, Adwaiya Srivastav, Debayan Gupta

Working Manuscript

Advisor: Prof. Sarvanan Vijayakumaran

Developed R1CS circuits for SHA3-256, SHAKE128, and SHAKE256 hash functions using the Arkworks library in Rust. Designed a step function for these and integrated them into the Latticefold codebase to create folding based IVC proofs.

Advisor: Prof. Debayan Gupta

Designed a protocol for enabling authentication of inputs in secrecy preserving two party computation protocols, specifically private set intersection. Extended the protocol to the setting of the authorized private set intersection and achieved improvements in runtime and communication overhead. Under submission. Currently designing a private set intersection protocol with input authentication based on KZG commitments with an aim to give a framework for reducing the overhead of PSI protocols to efficient commitment schemes. Developed a new variant of private set intersection on authenticated inputs called All-or-Nothing APSI.

Advisor: Prof. Debayan Gupta

Engineered features for radio frequency fingerprinting. Adapted it with an autoencoder-based anomaly detection model for detecting replay attacks against remote keyless entry systems in automobiles and achieved improved recall and accuracy.

Advisor: Prof. Debayan Gupta

Studied the Controller Area Network (CAN) bus in automotive vehicles and all known attacks against it. Investigated synthetic data generation using LSTMs and GANs to counter the unavailability of reliable data for training CAN bus intrusion detection systems.